Discussion: Polaris, AMD's 4th Gen GCN Architecture

Page 31 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Status
Not open for further replies.


If that is the upper limit it is very showing of the GloFlo 14nm.



I am willing to bet costs. Adding an additional 6pin or an 8pin would make the PCB bigger and more complex. They would also have to increase the phases on the PCB to handle higher power draws. That would also increase costs.

Would you buy it is it was $250-300 for the base 4GB and near $300+ for the 8GB?
 
6-pin cables are severely underrated, the 75W rating is silly low. Jonnyguru members have subjected 6-pin cables to > 200W for 8 hours straight without problems. In addition, you'll notice the 8-pin cable adds two ground wires yet doubles its rating.
 


and what psu was this?? i'm willing to bet it was not an oem pre-built pc psu. also doubt it was one of the common "average" psu's we see lots of people buying. we know that pc part makers have to consider these buyers as well as "enthusiasts" that buy better quality parts. somehow i don't think my dell oem 375w psu that is running an r9-270 fine would be able to handle 200w on that 6 pin connection very well. considering it also has a 105w q6600 cpu and multiple hdd's.

this card is an awesome upgrade at reference clocks for many people with pre-built pc's that happen to have a 6 pin psu. but not if it is trying to pull so much extra power above what it is supposed to.

same reasons manufacturers suggest a 650w psu when we know a quality 500w one would be good enough. they are considering less than quality units they know are in many many pc's around the world.
 
AMD/NVIDIA titles are just to get their name on there. Do people really think they're behind the scenes hurting each other while making the game?

AMD is better than NVIDIA at DX12, it's a fact. Don't you remember NVIDIA's claim, and why AMD recently benchmarked Ashes against their high end card? NVIDIA stated that Ashes was broken and in Alpha, now that it's released, what did they say? Absolutely nothing, I love it.
 


I don't think it would add that much to the price, but you will be proven correct or wrong once the custom designs arrive. And boy I am anticipating them with a lot of happiness. It appears the partners are going to go some-what wild with the designs.

I don't mind paying USD$300 if the card is cooler, quieter and better performing, while under ~220W.

Cheers!
 


come on now, it's not a secret than nvidia owns a game studio and those games do a lot better on bvidia cards. amd had a hand in ashes as well. even if they are not "officially" involved look at benchmarks. nvidia does very well in certain games and amd does very well in others. this is what folks don't like. why can't my card just do good in every game. there's more going on behind the scenes, for sure :)
 


I disagree. On multiple NVIDIA cards you will see performance not scaling as well going from DX11 to DX12. The R9 390x a year ago saw a 100% increase in FPS just by switching on DX12, where as the NVIDIA side got a 10-20% performance increase. What AMD and NVIDIA are doing to games are features you see in settings. I don't believe for a single second that they're putting money towards crippling eachothers overall game performance. I want proof of such a ridiculous claim, no proof so far. The "AMD/NVIDIA" titles shown in games are basically in Hollywood what the executive producers are, absolutely no involvment with the actual movie, but to get their name up there to increase sales/get sponsors.

If people want an even playing field, then blame NVIDIA. Blaming AMD means you want performance to decrease, no?
 
i'm not sure anymore what you're saying anymore?? i agree each brand makes games that favor their hardware. so you disagree with that when it's exactly what you're saying. can you clarify your point since clearly i missed it.
 
Sometimes it's hard to get my point across. But what I ment was that you said NVIDIA and AMD have their own involvement with games, and since NVIDIA is doing worse than AMD at DX12 in Ashes for instance, then it's instantly assumed that it's "AMD optimised", as in AMD cripple NVIDIA cards. That's what I understood your comment to mean, is that wrong?

I'm just saying that if AMD cards can match NVIDIA at DX11, NVIDIA should be able to match AMD at DX12, but they don't. If AMD are crippling NVIDIA, then why don't NVIDIA do the same thing? It doesn't make any sense to me, which is why I believe this is false, at least until there's proof of that.

My point was that NVIDIA must catch up to AMD, not the other way around, to favor NVIDIA.
 


That's a title heavily sponsored by AMD, why do you keep using it as your "for instance"?
 
they don't cripple nvidia, that's where your off base, i did not see anyone suggest such a thing. all gpu's are not made the same. amd has included pieces in their gpu's that take better advantage of async and other dx12 features. nvidia did not include these on maxwell and did not address it very well in pascal either. so ashes takes advantage of these hardware abilities the amd gpu has, thus better performance. it's not amd "crippling" nvidia. it is them taking advantage of everything they built into their gpu's and it just happens to hurt nvidia at the same time.

nvidia does the same thing. they take advantage of stuff built into their gpu's that amd does not have. physix for example. helps nvidia a lot and does nothing for amd. again not a deliberate crippling of amd in the game, rather taking advantage of what nvidia has to offer.

no big conspiracy or secret stuff going on. just each company taking advantage of what the have to show themselves in the best light. i don't see where you felt i was defending either or accusing either of this activity. this type of game development is common knowledge (well at least to those of us who actually pay attention to it all)
 
Yeah, and that makes perfect sense. It's only that NVIDIA were the ones that stated AMD ruined the performance of their cards. Maybe I'm just confusing that with what people say, although there are people who are on NVIDIA's side saying that they cripple their performance. I by mistake assumed this was just another case of that.
 


I'm talking more about the fundamental properties of the cable itself, not the PSU it is connected to. Karadjgne says 1 foot of 18 gauge wire at 12V DC can handle 5A of current. That is 60W. A 6-pin PCIe cable has 3 12V wires, making that theoretical maximum 180W. The 8-pin cable has the same amount of 12V wires anyway. Though if I am correct some 6-pin cables only actually use 2 of those 12V wires? Eh, it's confusing. That still right there is a rating of 120W for a 1 foot cable.
 


yah that is a mistake many make. fanboys in any case ignore the obvious and instead ramble conspiracy theories and hate to try and make a point. you are not on a fanboy forum. we have some here who ramble their nonsense, but there are plenty of use who actually know what we are talking about. and we love to correct them and show how uninformed they are. this is not reditt or facebook or wherever else those fools congregate and spread their nonsense. we are an actual tech site full of knowledgeable people. welcome to the forum and check the fanboy stuff at the door!!! 😀
 


In this case it does, AMD have been working very closely with Oxide to make sure that Ashes works well on GCN.
 


i know what you're saying but even if a cable can handle the draw, it does not mean the psu can provide it. :) isn't that something we tell noobs all the time around here? the cable/adapter does not create power, it only carries what the psu is capable of outputting 😛

so the card trying to pull so much power from a card rated so low only asks for trouble when it will be used on low wattage psu's that may not be able to handle the power needs, thus creating problems.
but you already know this 😀
 


Cables are "over stated" (or overrated) for their jobs because of power surges. If a cable doesn't support a high instant load (hard to explain in simple terms), then it could snap making arches or burn easily.

Electricity travels on the metal surface, so the electrical properties of the cables can be toyed with using different filaments or thicker cables. Each has pros and cons 😛

All in all, you never want to be at 90% theoretical capacity when talking high current.

Cheers!

EDIT: A human can be killed with 0.5A of current, no matter the amount of volts. Keep that in mind.
 


It is unclear what you mean when you say "electricity travels on the metal surface". There is no definition of what electricity is, it is a very generic term, so I do not understand what you are referring to here. If you are talking about the electrons in the wires, they do not flow just on the surface, they flow throughout the wire, not just on the surface. If you are talking about the electromagnetic energy, which performs the real work, it uses the wire as a medium but travels about the wire, on the outside of it basically through the air, and not through the wire.

But yeah I suppose those transient spikes are some things to consider. The RX 480 can easily reach 300W+ during a Furmark test for a brief instant, though its cables and everything stayed intact. Usually instantaneous stuff is too quick to cause any real damage.
 


The voltage does matter a lot, though. Since our bodies are ohmic resistors, it depends on the properties of the individual. Some people will have more restive skin and others have thinner skin with less resistance. The amount of current to kill somebody may remain about the same but the voltage matters a lot since voltage is what causes current. In general 12V is not enough to kill a human being, 40V is generally considered the point at which things become dangerous. But it depends a lot on your own skin's resistance, a person with good resistance may handle 60V fine. Then you have instances where people survive the high-voltage AC, mostly because high voltages are sometimes safer than medium voltages in that they can cause convulsions in your muscles and launch you away from the circuit.
 
The comment of how the electricity (electricity is defined as "electron movement") is about how different PSUs could use different type of cables. They usually ship with single copper lines instead of threaded ones (at least the ones that I've seen), so they have higher durability in terms of heat and wear&tear.

Cheers!

EDIT: What? No! Please go read about it in more detail. What kills is Amperage, not Voltage. If current manages to circulate for just a fraction of a second through some of your key organs, you will die. It is documented everywhere, so don't take my word for it.
 


You were the one just saying they might not be durable enough, though? And electricity is defined differently everywhere. Go in one dictionary to another, to a physics book... it's always different. It's quite obvious that the PCIe 6-pin cable is underrated when the 8-pin cable has double the rating and only adds 2 ground wires. There is still the same amount of 12V wires.

Also, voltage causes current. There cannot be current without voltage, because potential difference is what causes charge to flow.
 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom